Random Class Reference

Detailed Description

The generation of random bits is done with Mersenne Twister from U. of Hiroshima

http://en.wikipedia.org/wiki/Mersenne_twister http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html

Public Member Functions
	Random ()
	Constructor.
	~Random ()
	destructor
void	seed (const uint32_t s)
	seed with integer
uint32_t	seedTimer ()
	seed with time()
uint32_t	pint ()
	unsigned integer in [0,2^32-1]
uint32_t	pint_exc (const uint32_t n)
	unsigned integer in [0,n-1] for n < 2^32
uint32_t	pint_inc (const uint32_t n)
	unsigned integer in [0,n] for n < 2^32
uint32_t	pint_inc2 (const uint32_t n)
	integer in [0,n] for n < 2^32, (slow) integer based algorithm More...
int32_t	sint ()
	signed integer in [-2^31+1,2^31-1]; inlined for speed
int32_t	sint_inc (const int32_t n)
	integer in [-N, N], boundaries included
int32_t	sint_exc (const int32_t n)
	integer in [1-N, N-1], i.e. in ]-N,N[ with boundaries excluded
int32_t	int_range (const int32_t low, const int32_t high)
	random integer in [low, high] ( = low + pint_inc(high-low) )
uint32_t	pint_ratio (uint32_t n, const int ratio[])
	integer in [0 N], with probabilities given in ratio[] of size N, with sum(ratio)>0 More...
uint32_t	poisson (real E)
	integer k of probability distribution p(k,E) = exp(-E) * pow(E,k) / factorial(k) More...
uint32_t	poissonE (real EL)
	integer k of probability distribution p(k,E) = EL * pow(E,k) / factorial(k) More...
uint32_t	geometric (real p)
	number of successive unsuccessful trials, when success has probability p (result >= 0)
uint32_t	binomial (int n, real p)
	number of sucesses among n trials of probability p
bool	test (real const &p)
	true with probability (p), false with probability (1-p)
bool	test_not (real const &p)
	true with probability (1-p), false with probability (p)
bool	test_uint (uint32_t p)
	true with probability p / 2^32
bool	flip ()
	true or false with equal chance
int	sflip ()
	returns -1 or 1 with equal chance
int	sign_exc (const real a)
	this sign-function returns -1 or +1 randomly, if the argument is 0
real	preal ()
	positive real number in [0,1[, zero included
real	sreal ()
	positive real number in [0,n[ = n * preal() : deprecated, use preal() * n More...
real	preal_exc ()
	non-zero real number in ]0,1[
real	preal_exc (real n)
	non-zero real number in ]0,n[
real	real_range (real a, real b)
	real number uniformly in [a,b]
real	gauss ()
	random Gaussian number, following a normal law N(0,1) More...
void	gauss_pair (real &, real &)
	set two number, following a normal law N(0,1) More...
void	gauss_array (unsigned int n, real vec[])
	fill array vec with normal law N(0,1). More...
real	gauss_slow ()
	signed real number, following a normal law N(0,1), slower algorithm More...
real	exponential ()
	positive real x, according to distribution P(x) = exp(-x), expectancy = 1.0
real	exponential (const real E)
	positive real x, with distribution P(x) = exp(-x/E) / E : [ E = 1/Rate ]
template<typename T >
T	choice (const T &x, const T &y)
	uniform choice among the 2 values given: x,y
template<typename T >
T	choice (const T &x, const T &y, const T &z)
template<typename T >
T	choice (const T val[], int size)
template<typename T >
void	mix (T val[], int size)
	uniform shuffling of array T[]. More...

Member Function Documentation

real gauss

(

)

Signed real number, following a normal law N(0,1) using the polar rejection method (see Numerical Recipe)

void gauss_array	(	unsigned int	n,
		real	vec[]
	)

Fill n values in array vec[] with Gaussian ~ N(0,1).

void gauss_pair	(	real &	a,
		real &	b
	)

Signed real number, following a normal law N(0,1) using the polar rejection method (see Numerical Recipe)

real gauss_slow

(

)

this version uses cos() and sin() and is slower than gauss(). const real PI = 3.14159265358979323846264338327950288;

void mix ( T  val[], int  size  )

inline

Algorithm from knuth's The Art of Programming, Vol 2 chp. 3.4.2

uint32_t pint_inc2

(

const uint32_t

n

)

integer in [0,n] for n < 2^32

uint32_t pint_ratio	(	uint32_t	n,
		const int	ratio[]
	)

returns an integer in [0 n], with the ratios given in the array of ints

uint32_t poisson

(

real

E

)

expectation=E variance=E http://en.wikipedia.org/wiki/Poisson_distribution

This routine is very slow for large values of E. If E > 1000, the Poisson distribution can be approximate by a Gaussian distribution Poisson(E) ~ Gauss(E, E)

uint32_t poissonE

(

real

EL

)

This is equivalent to calling poisson(exp(-E)) The argument is EL = exp(-E) expectation=E variance=E (see wikipedia, Poisson Distribution)

real sreal ( )

inline

signed real number in ]-1,1[, boundaries excluded